Jacob Coxon asserts that he spent a full three years conducting research at both Anthropic and OpenAI.
雅各布·考克森(Jacob Coxon)聲稱,他曾在Anthropic與OpenAI從事研究工作凡三年之久。
On Tuesday, he observed on the social platform X that these two AI companies, far from prioritising safety, are in fact more intent on vying with each other—and with global competitors—to develop the most advanced models conceivable.
他於週二在社交平台X上指出,這兩家AI企業與其說是將重心置於安全之上,毋寧說是更熱衷於在研發盡可能最先進的模型上彼此較勁,並與全球競爭對手一爭高下。
OpenAI and Anthropic caused a considerable stir this summer, in that, roughly a week apart, they successively announced that their models had broken free of the confines of their testing environments and gained unauthorised access to real computer systems.
OpenAI與Anthropic於今夏掀起軒然大波,蓋二者相隔約一週,先後宣告其模型已掙脫測試環境之桎梏,並在未獲授權之下存取真實電腦系統。
Such announcements gave rise to widespread apprehension that the models might slip out of control and go on to carry out other, still more harmful tasks.
此等公告引發各界憂慮,恐模型失控,進而執行其他危害更甚之任務。
Both companies stated at the time that they were suspending certain evaluation procedures while introducing additional monitoring measures and safeguards.
兩家公司當時均稱,正暫停部分評估作業,同時採行更多監控舉措與防護機制。
The breakneck advancement of this technology has prompted a number of individuals in the United States and even world leaders to call for a more circumspect course of action.
該技術的迅猛發展,已促使美國若干人士乃至全球領袖紛紛呼籲採取更為審慎的行徑。
This week, the United Nations High Commissioner for Human Rights, Volker Türk, urged nations to "establish ironclad guarantees for the safety and security of AI before it is too late."
聯合國人權事務高級專員福爾克爾·蒂爾克(Volker Türk)本週敦促各國,「在為時已晚之前,對AI的安全與保障建立鐵一般的保證」。
In a social media post, Coxson stated bluntly that Anthropic and its chief rival OpenAI "are racing headlong toward self-improving superintelligence, gambling with our lives."
考克森於社群媒體貼文中直言,Anthropic及其主要競爭對手OpenAI「正直奔自我改進的超智慧,拿我們的生命賭博」。
He warned that several individuals engaged in AI development believe this technology may endanger human life before the end of this decade.
他警告,若干投身AI開發之人認為,這項技術或將於本十年結束前危及人類生命。
"Never underestimate the power contained within this technology," he continued.
「切勿低估這項技術所蘊含的力量,」他續道。
"These will swiftly metamorphose into superhuman systems, capable of hacking into anything, utterly upending any domain overnight, and seizing genuine power and resources.
「這些旋即將蛻變為超人系統,得以駭入任何事物、於一夕之間徹底顛覆任何領域,並攫取真正的權力與資源。
We have already witnessed advances across various fields, and the progress has not abated in the slightest."
我們已目睹各領域的進展,而進展未曾稍歇。」
Coxson is by no means the first person within the AI industry to voice such concerns publicly.
考克森絕非率先公開表達此類疑慮的AI業內人士。
In recent years, both Anthropic and OpenAI have experienced high-profile resignations prompted by safety concerns.
近年來,Anthropic與OpenAI均曾發生因安全顧慮而觸發的備受矚目之辭職事件。
Two current Anthropic employees also responded to Coxson's post, expressing their agreement.
兩名Anthropic現任員工亦就考克森的貼文作出回應,並表達認同。
Bernie Sanders, the independent United States Senator from Vermont, has urged the establishment of safeguards and regulatory standards for artificial intelligence; he shares the concerns held by Coxon, and has stated that he will soon introduce legislation to pause the development of artificial intelligence while prohibiting research into superintelligence.
佛蒙特州無黨籍聯邦參議員伯尼·桑德斯(Bernie Sanders)曾敦促針對人工智慧制定保障機制與監管規範;他認同考克森所抱持的疑慮,並表示將於近期提出立法,暫停人工智慧的開發,同時禁止超智慧之研發。
AI companies themselves have at times been at pains to play up the threat that the technology poses to humanity, while sceptics contend that this is nothing more than one of the promotional ploys by which they make their own products appear omnipotent.
AI企業本身亦時而著力渲染該技術對人類所構成之威脅,而抱持懷疑態度者則認為,此舉無非是令自家產品顯得無所不能的宣傳伎倆之一。
Coxon points out that the various fears he outlined in his post are by no means a "marketing gimmick".
考克森指出,其於貼文中所概述的種種憂懼,絕非「行銷噱頭」。
Ever since its founders departed from OpenAI in 2021 to strike out on their own and establish this startup, Anthropic has long positioned itself as the more responsible and safety-conscious among the leading AI companies.
自其創辦人於2021年離開OpenAI、另立門戶創辦這家新創公司以來,Anthropic便長期自我定位為領先AI企業中更負責任、更重視安全的一家。
The company stated recently that it is taking action to "prioritise safety when safety and speed come into conflict".
該公司近日表示,正採取行動「當安全與速度相衝突時,優先考慮安全」。
Anthropic and OpenAI are both stepping up preparations for highly anticipated initial public offerings (IPOs), even as they find themselves mired in an intensely heated rivalry with one another.
Anthropic與OpenAI雙雙為眾所矚目的首次公開募股(IPO)加緊籌備,並深陷於彼此之間劍拔弩張的激烈競爭之中。
At the same time, each of them bears a mission — to outpace Chinese AI companies in the race to develop — and the Trump administration has remained eager to come out on top in this contest.
與此同時,兩者各自肩負著一項使命——在開發進度上超越中國AI公司——而川普政府始終熱衷於在這場競賽中拔得頭籌。